Microcontroller Selection Questions and Answers

Answer: C

Correct answer: whether all six interfaces can be routed to pins at once

A 32-pin package has perhaps 26 usable I/O lines, and the peripheral count in the summary table is the number of instances on the die, not the number that can be brought out simultaneously. The multiplexing table decides it, and a part advertising more peripherals than its package can expose is entirely normal. The core speed and flash size are secondary until that is settled.

Answer: D

Where two parts are otherwise equal, the one the team already knows saves real weeks of learning and toolchain setup. Allowing it to override a peripheral mismatch or a temperature grade, however, is how the wrong part gets chosen.

Answer: A

Correct answer: 3.2 mV

\[ \frac{3.3\ \mathrm{V}}{2^{10}} = \frac{3.3}{1024} = 3.22\ \mathrm{mV}. \] That is the size of one least significant bit. It is a theoretical figure: noise, reference stability and the converter’s own linearity mean the effective number of bits is usually a little lower, so a 10-bit reading is rarely good to one count.

Answer: B

A TCP stack, a graphics buffer or a file system each want kilobytes that nobody budgets for at the outset, and RAM is the resource that cannot be traded for anything else. Twenty-five per cent spare is a common and defensible provision.

Answer: C

Correct answer: to weigh that licence in the total cost of the selection

A tool licence is a real and recurring cost of using the part, so it belongs in the comparison alongside the unit price, the development effort and the risk. Sometimes it is worth paying and sometimes another part with a free toolchain wins on the total; that is a calculation rather than a matter of principle. Using an unlicensed compiler for production is not an option.

Answer: D

Correct answer: performs sustained floating point arithmetic in a timed loop

A hardware unit earns its cost where floating point operations happen often and to a deadline - a control loop, a filter, a transform - because software emulation takes tens or hundreds of cycles for what the unit does in one. Floating point confined to initialisation costs nothing worth optimising, and the language used is irrelevant since the compiler emits the same operations either way.

Answer: A

The certification asks what is known about how the component behaves when it fails and what process produced it, and only the vendor can answer. Sealing is the enclosure's business and redundancy is a design decision rather than a part attribute.

Answer: B

Popularity correlates with support, tooling and supply, not with performance; the fastest part in a class is frequently a recent and thinly supported one. Weighing the two against each other is the whole of the decision.

Answer: C

The matrix records what mattered and which alternatives nearly qualified, so the search starts from a shortlist rather than from nothing. The schematic and the code say what was built but not why, which is the question a replacement has to answer.

Answer: D

Getting one character wrong brings a commercial-grade part in the wrong package, which is a surprisingly common and expensive mistake. Lot and date codes are marked separately on the package rather than encoded in the ordering part number.